4
$\begingroup$

This feels like a very natural question, but I am struggling to find a reference, proof, or counter-example.

Consider a simple graph $G$, and say two vertices are conjugate if there exists an automorphism of $G$ that switches them. It is not hard to show that if $v,w$ are conjugate than the induced subgraphs on $G\setminus v$ and $G\setminus w$ are isomorphic.

But is the converse true? In my context of interest, multiple edges between vertices are allowed and all edges are undirected (although I don't anticipate that this affects the answer).

An equivalent statement is that for any isomorphism $\phi$ from $G\setminus \{v\}$ to $G\setminus \{w\}$ and vertex $u$ in $G\setminus \{v\}$, $\phi(u)$ and $u$ have the same degree in the original graph $G$.

$\endgroup$
4
  • 2
    $\begingroup$ My intuition, guided by the Graph Reconstruction Conjecture, is that the answer to your question is "No." In the study of the GRC, the induced graph caused by deleting a vertex is the "card" corresponding to that vertex and the GRC claims that the full set of vertex-deleted cards determines the graph. The GRC is still open after 67 years and I think your question is a weaker question in relation to the GRC. So I would guess that there is a small (10 vertex or less?) counterexample. (But I do not know one.) $\endgroup$ Commented Jan 28 at 0:11
  • $\begingroup$ I see, thank you for the response. If the graph was a tree, would it be easier to prove true? $\endgroup$
    – Terence C
    Commented Jan 28 at 0:28
  • 1
    $\begingroup$ I would indeed start by looking at trees. Or regular graphs. $\endgroup$ Commented Jan 28 at 0:31
  • 2
    $\begingroup$ @KenW.Smith your intuition was correct: math.stackexchange.com/questions/4852344/… $\endgroup$
    – Terence C
    Commented Jan 28 at 1:19

1 Answer 1

2
$\begingroup$

Pick any vertex transitive graph $G$ (of degree at least 2) that has vertices $u,v \in V(G)$ such that no automorphism of $G$ swaps $u$ and $v$ (in other words, $G$ is vertex transitive but not generously transitive). Create $G'$ by adding vertices $u'$ and $v'$ to $G$ and making them adjacent to $u$ and $v$ respectively. Then clearly $G'-u'$ and $G'-v'$ are isomorphic but there is no automorphism that maps $u'$ to $v'$ since its restriction to $V(G)$ would have to be an automorphism of $G$ that swaps $u$ and $v$.

Cayley graphs with connection sets that are unions of conjugacy classes (e.g. Cayley graphs for abelian groups) are generously transitive, but I suspect that if you pick a random one without this property then it will very likely not be generously transitive (of course it will be vertex transitive).

$\endgroup$
1

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .